FDA's fast track program may be the most successful of FDA's user fee-era regulatory initiatives aimed at speeding new drug reviews, an analysis by the Tufts Center for the Study of Drug Development suggests.

"The regulatory characteristic with the largest difference between its single-cycle approval rate and that for all approvals was for compounds approved through the fast track program," Tufts Center for the Study of Drug Development researchers Joseph DiMasi and Laura Faden report in a study published in the Drug Information Journal (vol. 43, 2009).

"More than three-quarters of the fast track drugs were approved after just a single review cycle, compared to approximately half of all new drugs and biologics," DiMasi and Faden found.

The researchers analyzed information on 298 approved NMEs and therapeutically significant BLAs that had been submitted in fiscal years 1996 through 2006 and approved through fiscal year 2006. Forty-six applications had fast track status.

Fast Track Could Be Surrogate For Innovation

The fast track program, designed to "facilitate development and expedite reviews," is open to new products that aim to treat serious or life-threatening conditions and new drugs with the potential to address unmet medical conditions. The program allows for rolling submission of reviewable units of an application. FDA's fast track guidance, originally published in 1998, also recommends a framework of meetings and interactions with FDA throughout the development cycle. Sponsors can request fast track status any time after IND submission.

The criteria for gaining fast track status provide some clues as to why this regulatory program correlates with first-cycle approval. First, the products that qualify are more likely to be novel drugs for conditions with a real demand for therapy - and true medical advances have an easier time passing through FDA. Secondly, the program itself offers more early communication and opportunities for advice from the regulators throughout development.

The Tufts analysis "did not test hypotheses about reasons behind the results for fast track," DiMasi, who is the CSDD's director of economic analysis, noted in an email interview. However, he continued, "one might speculate that, because these applications are for drugs to treat serious or life-threatening conditions, the FDA may devote more resources toward advising on their development and in reviewing the applications."

FDA's Efforts Working As Planned

While drug safety has eclipsed review speed as FDA's top public priority in recent years, the agency has continued to pursue efforts to improve the drug review process - and thereby increase first-cycle approvals and faster drug reviews - primarily under the rubric of Good Review Management Principles.

The Tufts CSDD findings appear to support many of FDA's tactics to enable faster drug reviews. "Drugs approved under regulatory programs designed to speed approval, such as the accelerated approval and fast track programs … had much lower than average approval time," DiMasi and Faden wrote.

The study also identified other regulatory variables that support the association of early sponsor-FDA interaction and swifter reviews. In promoting GRMPs, FDA has stressed the value of laying a solid foundation of understanding and communication with FDA early in the development process (¹ Pharmaceutical Approvals Monthly July 2008, p. 18).

Other markers of early interaction with FDA, like end-of-Phase II and pre-NDA meetings, were associated with higher rates of first-cycle approval in the Tufts CSDD study. The single-cycle approval rate for the milestone meetings was slightly shy of two-thirds; the single-cycle approval rate for all NDAs/BLAs in the study was 51 percent (see ² graph).

The notable effect of milestone meetings bears out the conclusions of an earlier analysis of FDA review performance by Booz Allen Hamilton done under the prior PDUFA reauthorization round. The consulting firm analyzed 77 NMEs submitted in fiscal years 2002-2004, and found a notable effect of EOP2 meetings in particular - a 52 percent rate of first-cycle approval, vs. 29 percent for products lacking the mid-clinical development interaction (³ Pharmaceutical Approvals Monthly February 2006, p. 3).

BAH took a second look at the data, noting in a July 16, 2008 report that "sponsors continue to take advantage of the pre-submission meeting opportunities," with "a significant increase in the number of applications that had an EOP2 and/or Pre-NDA/BLA meeting from FY2002 to FY2007."

In fact, the role of EOP2 meetings was so clear that FDA entered into the current PDUFA round requesting additional funding because of the industry demand for such meetings (⁴ Also see "PDUFA IV: FDA Stresses Burden Of Industry Meetings In User Fee Discussions" - Pink Sheet, 21 November, 2005., p. 3).

To Facilitate And Expedite

While EOP2 meetings have grabbed headlines, the fast track program has had a low profile - especially after BAH found little effect from two PDUFA III pilot programs aimed to improve fast track.

In October 2003, FDA introduced the Continuous Marketing Application Pilot 1, which formalized FDA's commitment to review early submissions of reviewable units of eligible fast track applications. Traditional fast track status allows for rolling submissions, but makes no commitment for early review by FDA. CMA Pilot 2 followed with a plan to ensure frequent interaction during the IND phase, a stiffening of the original guidance's recommendation of a schedule of interactions.

BAH assessed the pilots, and, in a report published May 10, 2006, found that the CMA programs did not have an impact on application quality. FDA scrapped the pilot programs, saying that the added benefits did not justify the costs to the agency (⁵ Also see "PDUFA Pilots: Risk/Benefit Plan Takes Off As “Continuous” Applications Stop" - Pink Sheet, 29 January, 2007., p. 15).

While the consultants found a higher first-cycle approval rate for fast-track than non-fast track products (58 percent vs. 42 percent), BAH also questioned the fast track program overall in the 2006 report. "In general … it is not clear whether an early review of Rolling Submissions occurs consistently," the consultants wrote. "As with the pilot products, the comparison cohort [of fast-track products] shows that the impact of early review on first-cycle approvals is still unclear."

The Tufts CSDD team, on the other hand, appears to have found some clarity on the impact of early review. "The odds of a multiple cycle review for compounds approved under the fast track development program were less than one-third the odds for compounds approved without fast track status," the study reports.

Among applications that were not approved on the first cycle, the Tufts CSDD researchers looked at sponsor response times to the FDA action. "Fast track drugs had relatively quicker sponsor response times, while drugs that had advisory committee meetings or were approved under the accelerated approval regulations had somewhat longer than average sponsor response times (suggesting that these drugs had more difficult problems to resolve in cases where the drug was not approved on the first cycle)."

Fast track and accelerated approval drugs were also both associated with shorter approval phase times, the study found. The mean time to approval for all drugs was 15.5 months, the study says. The mean approval phase for fast track drugs was 9 months; the mean for accelerated approval products was 11.6 months (see chart for ⁶ more approval phase and other data for selected regulatory variables ).

Adding A Multivariate Twist To Sparse Literature

As the article notes, "the literature on FDA review cycles is sparse." The BAH report is the most recent major evaluation, and it relied on relatively recent data: submissions during fiscal years 2002-2004, later extended to include FY 2005-2007. The Tufts researchers, however, describe their dataset as "much larger": all NMEs and significant BLAs submitted during FY 1996-2006 and approved through FY'06.

The BAH analysis "considers many factors that may be associated with multiple review cycles, but the authors considered the sample to small for statistical analysis," according to the Drug Information Journal article. "Only descriptive statistics are provided for single and multiple cycle review rates by numerous characteristics."

In addition to descriptive statistics, the Tufts CSDD researchers used multivariate statistical techniques to "establish associations between various drug, sponsor, and regulatory characteristics and the likelihood of longer or shorter approval times."

DiMasi elaborated that the advantage of the CSDD multivariate statistical approach "is that you can have greater confidence that the differences in results by factor that you observe are real if they are found to be statistically significant (and not due to chance or other confounding factors)."

But the value isn't just in analyzing retrospective data sets - the findings have practical implications for drug developers. "Statistical models such as those presented here can help improve the precision with which drug developers and regulators predict approval times and assess the effectiveness of regulatory programs designed to speed the drug approval process," the Drug Information Journal article states.

Priority Reviews Propel Rising Approval Times

"Although the entire approval process has, on average, shortened during the user fee era, there are still variations in approval times that are worth considering," the authors wrote.

Most analyses of review times must deal, if only indirectly, with the Vioxx question: did FDA become more conservative following the September 2004 withdrawal of the Merck COX-2 inhibitor? The Tufts CSDD data set was too early to address the Vioxx effect directly, but the authors did find a slow-down in average approval times predating the high-profile withdrawal - and discovered that the increasing approval time was due to priority review products.

The Vioxx impact came late in the Tufts review period, but DiMasi and Faden acknowledge the perception of that factor: "Some industry and other observers have speculated that approval times have become lengthier in recent years because of heightened sensitivity to safety issues, particularly in the years since the [COX-2] inhibitor rofecoxib (Vioxx) was withdrawn."

Even before the Vioxx withdrawal, the study identified a trend toward slower approval times.

The average approval times for the applications submitted from FY'96 to FY'03 were divided into two three-year subperiods: (FY'96-FY'99, with 147 applications, and FY'00-'03, with 114), with "both higher mean and median approval times" in the later cohort. The earlier submission period had a mean 15.4 month time to approval, which rose to 18.1 months for the later submission group.

"The longer approval times for the more recent submission period were primarily [caused] by longer approval times for priority drugs," the researchers state. The submissions studied were close to evenly divided between priority applications (144 submissions total) and standard review products (154).

The Tufts analysis looked at "approval phase time" - meaning the time from NDA submission to approval, including sponsor response time.

The mean approval phase for priority review applications submitted FY'00-'03 was 14.3 months, up from 10.4 months for the earlier FY'96-'99 cohort. Standard review applications showed a less steep increase, from a mean approval phase time of 19.3 months in the late '90s submission group, edging up to 20.8 months in the early '00s cohort.

"A more discriminating regression analysis showed that the longer approval phase times were more specifically driven by compounds that had priority ratings and were subject to multiple review cycles."

The authors also observe that approval phase times were, as expected, substantially lower for priority compared with standard review applications - but add that, "interestingly, there was more variability … in approval times for priority drugs."

Therapeutic Category Matters

DiMasi and Faden, looking at the data by therapeutic category, found that "most categories had averages [of approval phase time] that were within a narrow band." There were, however, "substantial differences" in the mean approval times for antineoplastics and anti-infectives at the low end and CNS and respiratory drugs at the high end. To illustrate the divide, the authors note that "the mean and median for the respiratory class were approximately twice as high as those for the antineoplastic class." (See chart for ⁷ approval statistics by therapeutic category.)

"More than two-thirds of the anti-infective and antineoplastic drugs were approved after only one review cycle, while only one quarter of the gastrointestinal drugs and 17 percent of the CNS approvals were approved in one cycle," the study reports.

"Therapeutic class had a statistically significant effect on the odds of multiple-cycle reviews," the authors report. However, they continue, "we failed to find an effect on approval phase times for a given number of review cycles. … It could be the case that differences in average approval phase times by therapeutic class were driven primarily by the differences in the [proportion] of compounds in the classes that had multiple cycle reviews."

That's as far as the article goes, however. The authors caution that "the data available do not allow us to distinguish between inherent variation in the degree of uncertainty or difficulty involved in testing drugs in different therapeutic classes and variation in either regulatory stringency or efficiency across FDA reviewing divisions."

Biologics vs. Drugs

Another common focus when evaluating FDA review performance is to compare biologic products, which can be more complex and historically have taken longer to review, with new molecular entities.

However, DiMasi and Faden found that new biologic products, on average, took only a few weeks longer to approve than NMEs - a mean of 16.2 months from submission to approval for the 57 BLAs in the study, compared with a mean of 15.3 months for the 241 NDAs, according to the study. Similarly, CDER and CBER had roughly equivalent approval times for biologics reviewed by the different centers. CDER took over review of therapeutic biologic products from CBER in late 2003; CBER still reviews vaccines and certain other biologic applications.

New molecular entities "had a somewhat higher first-cycle approval rate than did [all biologics] (52.7 percent for NDAs vs. 43.9 percent for BLAs)," the study found. "The lower first-cycle approval rate for BLA submissions was associated with higher shares for three- and four-cycle reviews."

The study does drill down to compare CDER and CBER review statistics on biologics (although they only had a small number to work with), finding that CBER-approved compounds "had an even lower single-cycle approval rate relative to CDER (36.6 percent vs. 53.3 percent) than did BLAs relative to NDAs."

Despite a higher first-cycle approval rate at CDER, the Tufts CSDD researchers found that CBER-reviewed products had a shorter time to approval. Other things being equal, the average time of a first-cycle approval for a biologic was 18 percent higher in CDER than in CBER.

Overall, the study found that "compounds approved through CBER that had multiple review cycles and compounds approved under the accelerated approval regulations were associated with shorter approval phase times."

- Bridget Silverman (b.silverman@elsevier.com)

Fast Track Speeds Approval, Tufts Finds; Therapeutic Category Also Key

More from Archive

More from Pink Sheet